We undertook a prospective evaluation of 4 methods for the detection of Pneumocystis carinii in clinical specimens and compared an indirect immunofluorescence assay (IFA) (Diagnostics Pasteur), and a fluorescent whitening agent (FWA) (Blankophor BA 267%, Bayer, Australia) with our standard methenamine silver (MeAg) and toluidine blue O (TB) stains. Two hundred and two specimens were received from 162 patients (133 HIV infected, 19 heart or heart-lung transplant recipients, and 10 “miscellaneous”). The specimens consisted of 132 induced sputa, 56 bronchoalveolar lavage specimens, 10 fine needle aspiration lung biopsies, and 4 pleural fluid specimens. P. carinii was detected in 44 (22%) of the specimens. The sensitivities for the detection of P. carinii pneumonia were IFA: 92% (95% Cl, 83–100%), FWA: 57% (95% Cl, 41–73%), MeAg: 54% (95% Cl, 38–70%), and TB: 49% (95% Cl, 33–65%). Discordant results were greatest in specimens from patients who were receiving specific anti-P. carinii prophylaxis, or who had received treatment for several days prior to sampling. IFA was the most sensitive test and relatively easy to perform. IFA was also the most expensive test. We found the FWA method a useful screening test as it is cheap and quick to perform. However, it is less sensitive than IFA, which should be performed on the negative specimens. With the increasing use of specific anti-P. carinii prophylaxis in HIV-infected patients, methods more specific and sensitive than MeAg and TB stains are required. We have found IFA to improve significantly the rate of detection of P. carinii in this patient group.